1. Field of the Invention
The present invention relates to an information processing apparatus that can recognize user actions.
2. Description of the Related Art
A device that can detect a hand area (i.e., an area that includes a user hand) from an image captured by a visible light camera or an infrared ray camera and can recognize a gesture operation or an operation performed on a displayed user interface based on the movement and the position of the detected area is becoming widely used.
As discussed in Japanese Patent Application Laid-Open No. 2009-104297, a difference between an image obtained by capturing a user hand that is present in the background and a background image obtained by capturing only the background can be extracted using the background subtraction technique, which is a technique capable of detecting a hand that performs a gesture operation. Using the above-mentioned method is useful to detect the hand area and determine an operation based on the shape or the movement of the detected hand area.
Further, a method for managing a background image behind a substantial object discussed in Japanese Patent Application Laid-Open No. 2010-92224 includes dividing an imaging range into a plurality of partial areas and discriminating a moving body area that includes a moving body (e.g., a human object) from a background area that does not include any moving body for each partial area of an input image. The method further includes combining the latest background areas obtained from respective partial areas in such a way as to obtain the latest background image by updating the background image that does not include any moving body in the entire imaging range.
However, to accurately detect a recognition target (e.g., a user hand), in the above-discussed methods it is important that an object included in the background image is stationary. For example, if the recognition target moves together with an object other than the recognition target in the background area, an operation originated by the movement of the recognition target may not be accurately detected.